Short Time Vibration Analysis and Parameterisation as a Tool for
Machine Prototypes TestingVibrations in Physical Systems 2020, 31,
2020112 (1 of 10)
Short Time Vibration Analysis and Parameterisation as a Tool
for Machine Prototypes Testing
3 Piotrowo St, 60-965 Poznan,
[email protected]
Abstract
The paper outlines the idea of using the short-time vibration
signal processing for testing machines of a complex
design, particularly the machines consisting of many subassemblies
with a non-stationary (e.g. cyclic) operation modes. The method
presented consists in graphic representation of vibrations of
subassemblies in the form of
a trace on a plane. It is created by associating instantaneous rms
values of vibration acceleration and
instantaneous Rice frequency values obtained as a result of
short-time signal processing. On the basis of the shape,
orientation of the created trajectory and / or dispersion /
concentration of points on the Rf – arms plane, the
type of non-stationarity of generated vibrations can be identified.
The presented methodology can be used for
testing machine prototypes. The results in the proposed form can be
helpful to determine the type of vibration reduction systems for
individual machine subassemblies. It is also possible to detect
subassemblies for which
an increase in machine capacity results in an increase in the level
of generated vibrations. The location of the
averaged values of measures obtained for a new machine on the Rf –
arms plane can be a reference point for further monitoring of
machine vibration and for detection of damage or malfunction of its
subassemblies.
Keywords: machine vibration, non-stationarity identification, short
time signal analysis
1. Introduction
A group of machines can be distinguished for which the methodology
of the assessment
of their technical condition on the basis of vibration
measurements, specified in standards
(e.g. [1, 2]) is not adequate. This may result from the range of
rotation speeds or from the
variety and specifics of working movements (not necessarily
rotational) performed by
machine subassemblies. Machines with a complex design and
additionally with many
drive units (e.g. electric motors, pneumatic cylinders) are used,
among others in the food,
textile and tobacco industry. In this case, the evaluation of
machine vibrations based on
the aforementioned standards is not justified. Therefore, the
problem arises how to run
prototype tests and how to assess the technical condition of this
class of machines during
operation. Methods are sought and developed that allow detection of
machine
malfunctions. One of possible solutions is to use the change of
poles location – natural
frequency and damping ratio in order to detect and localize the
machine malfunctions [3].
In this paper the author proposes a slightly different approach to
running vibration tests
of machines of a complex design. It uses a graphic representation
of results of short-time
processing of vibration signals recorded at representative
measuring points located on
individual subassemblies of the machine. This method involves
synchronous
determination and visualization of the location of instantaneous
rms values of vibration
accelerations arms(τ) and Rice frequencies Rf (τ) on the Rf – arms
plane. This approach allows
one to determine the specifics of the operation of subassemblies
e.g. during prototype
Vibrations in Physical Systems 2020, 31, 2020112 (2 of 10)
testing. On the other hand, the observation of the change of the
position on the Rf – arms
plane of the averaged Rf ( ) and arms ( ) values during operation
time can be the basis
for detection of malfunctions of subassemblies.
The paper presents an outline of the proposed alternative
methodology and sample
results obtained during vibration tests of the prototype of a
horizontal tray former. The
purpose of this research (in the short term aspect) was to obtain
data that can be helpful
for optimizing the construction of the prototype of the former and
for selection of vibration
elimination systems. Test results (in the averaged values aspect)
can be treated as reference
values for further monitoring of the technical condition of the
machine during operation.
2. The test object
The FTHT6 former (Fig. 1) is a modular machine used for the
erection of cardboard trays.
This machine has a cyclical operation mode. It consists of several
cooperating
subassemblies which make various working movements: rotations of
variable speed as
well as reciprocating and complex movements. Some phases of forming
and bonding of
a cardboard have an additional impact nature (e.g. bending,
pressure during the gluing
phase).
Figure 1. FTHT 6 former and vibration accelerations recorded at
representative
measuring points p.1 – p.10 (one machine operation cycle is
presented) [4]
The research was carried out for the rated cardboard formation
speed, i.e. 25 cycles
per minute. Vibration accelerations were recorded at 10
representative points located on
the former’s prototype:
2 – cardboard manipulator rotary beam (upper),
3 – servo drive beam,
4 – punch gearmotor base,
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
p1 p2 p3 p4 p5 p6 p7 p8 p9 p10
20 10
Vibrations in Physical Systems 2020, 31, 2020112 (3 of 10)
7 – transverse beam for cardboard slide (at the end),
8 – the end of the guide (channel bar),
9 – cardboard bender,
10 – pneumatic clamp (pressure).
Their locations were determined on the basis of kinematic analysis
of the machine and
suggestions from the construction team. Figure 1 also shows a
fragment of recorded
vibration accelerations covering one cycle of operation. The
variety of dynamic
interactions of individual subassemblies is visible.
3. Methodology
Vibration accelerations at representative points were recorded
synchronously due to the
specificity of dynamic interactions, the complexity of the FTHT6
former and its high
capacity. For the vibration tests of the prototype of the former
eight IMI® (type 627A01,
603C01) and two MMF (type KD10) accelerometers were used. The
signals from the ICP
(IMI) standard transducers were recorded directly by an
eight-channel recording unit
TEAC LX-10. The signals from the charge transducers (KD10) were
pre-conditioned in
a preamplifier (Nexus B&K) and recorded by a two-channel data
acquisition module
VibDAQ 2+. The way the transducers were mounted and the
analog-to-digital conversion
parameters (sampling frequency 6 kHz, 24 bit quantization) ensured
a linear vibration
measurement with an appropriate dynamics in the frequency band to a
minimum of 3 kHz.
The signals were analysed and parameterised in an application
developed in the
DASYLab (Data Acquisition System Laboratory) environment. The
digital signal
processing included: signal windowing, STFT (Short Time Fourier
Transform) analysis
[5, pp 46-52] as well as, basing on time-frequency (TF) maps, the
determination of
instantaneous rms values of vibration accelerations arms(τ),
instantaneous Rice frequencies
Rf (τ) and creating a display of results on the Rf – arms plane.
Further data synthesis and
parameterization were enabled.
A rectangular time window with a size of 512 samples was used to
segment the data
in the process of signal processing. As a result, short-time,
0.085-second-long signal
sequences were obtained. The time window shift τ in the STFT
analysis was equal to the
window size (overlapping = 0). TF maps were the basis for further
parameterisation of
vibration accelerations. An example of such a map (2D and 3D) is
shown in Figure 2. The
sonogram shows a broadband impulse and subsequent monoharmonic
damped vibrations
of the bender (measurement point no. 9) with a frequency of approx.
75 Hz.
Vibrations in Physical Systems 2020, 31, 2020112 (4 of 10)
0
1
Hz
250 500 750 1000 1250 1500 1750 2000 2250 2500 2750
Y/t Chart 0
500
450
400
350
300
250
200
150
100
50
0
Figure 2. An example of STFT analysis covering four operation
cycles of the former
(accelerations recorded at the measurement point no. 9 – cardboard
bender) [4]
It should be mentioned that Rice frequency Rf describes the average
frequency of the
analysed process [6]. It is used, among others, in the diagnostics
of rolling bearings, for
the detection of damage of valves in internal combustion engines
[7] and cavitation [8].
The Rice frequency can be determined based on power density spectra
[9 pp 84-85].
Bearing this in mind, the instantaneous Rice frequency values Rf
(τ) can be obtained as
a result of post-processing of subsequent short-time spectra with
the following formula
[10]:
(1)
where:
Xw(f,τ) – sonogram obtained as a result of STFT of a(t)
signal,
f – frequency,
τ – shift of the analysis time window.
The properties of the obtained Rf (τ) characteristics and methods
of its further
parameterisation are shown in a monograph by the author [11].
Bearing in mind the
Parseval's theorem [12, p. 134] instantaneous arms(τ) values can be
easily determined by
synthesizing short-time spectra. A similar method of determination
(calculation) of the
rms value of the signal from the spectrum was used in another work
[13]. As part of the
research, averaged values of the measures f, and rms, for 20 s of
signal covering 10
complete machine cycles were also determined.
It should be added that the determination of the above measures
based on
post-processing of TF maps gives a possibility of conducting a
selective analysis in any
0
1
frequency f [kHz]
Y/t Chart 0
0 10 20 30 40 50 60 70 80 90
Vibrations in Physical Systems 2020, 31, 2020112 (5 of 10)
frequency range. This is done by taking into account in the
processing of components of
a TF map located between arbitrarily set cutoff frequencies.
4. Results
An idea of interpreting the results of short-time parameterisation
of vibrations is shown in
Figure 3. It can be intuitively determined whether the signal is
stationary or non-stationary
in the amplitude and / or frequency terms. The proposed method can
be an alternative to
the methods of detecting signal instability described by Plazenet
et. al [14].
Figure 3. The idea of interpreting the results of short-time
parameterisation of vibrations;
a representation of various vibration signal types on the Rf – arms
plane:
A-N non-stationary amplitude; F-N non-stationary frequency, Q-S
quasi-stationary
signal, AF-F non-stationary signal in amplitude and frequency
terms
in st
an ti
n o
u s
a rm
Vibrations in Physical Systems 2020, 31, 2020112 (6 of 10)
Figure 4. Results of acceleration parameterisation on the Rf – arms
plane,
averaged values for 20 cycles of machine operation [4]
The results shown in Figure 4 allow one to estimate which
subassemblies are
characterized by the highest vibroactivity. The results also allow
one to determine the
average vibration frequency. Three subassemblies can be pointed
out: cardboard bender
(no. 9) pneumatic clamp (no. 10) and guide (no. 8) for which
reduction of vibration levels
may be considered. Signal parameterisation results (Table 1) are
helpful in this aspect. The
results are supplemented with acceleration peak values apeak and
crest factors k.
Table 1. The results of parameterization of vibration accelerations
(frequency band up to
2.5 kHz) for 10 operation cycles of the FTHT6 former (approx. 20
s)
Measure
[unit]
1 2 3 4 5 6 7 8 9 10
rms [m/s2] 1.3 4.6 1.7 1.5 5.4 2.0 1.9 7.4 38.1 12.5
apeak [m/s2] 15.8 59.1 20.9 53.2 45.5 16.1 46.2 168.6 810.0
451.9
k [-] 12.5 12.7 12.0 36.3 8.4 8.0 24.0 22.8 21.3 36.2
f [Hz] 1130 1182 1365 1409 1008 806 1358 529 241 680
An example of a graphic representation of the short-time
parameterisation of vibration
accelerations on the Rf – arms plane is shown in Figure 5.
0
5
10
15
20
25
30
35
40
Vibrations in Physical Systems 2020, 31, 2020112 (7 of 10)
a)
b)
c)
Figure 5. Example results of parametrisation of accelerations on
the Rf – arms plane;
a) punch gearmotor base b) punch c) cardboard bender [4]
0
5
10
15
20
25
30
a rm
s [m
/s 2 ]
Rf [Hz]
a rm
s [m
/s 2 ]
Rf [Hz]
a rm
s [m
/s 2 ]
Rf [Hz]
Vibrations in Physical Systems 2020, 31, 2020112 (8 of 10)
A gearmotor is a device consisting of an electric motor and a
gearing. The specificity
of vibrations generated by the gearmotor is shown in Figure 5 a.
Non-stationarity in both
the amplitude and frequency terms is visible there. The cyclical
start of this subassembly
causes an increase in instantaneous arms(τ) values as well as
instantaneous Rf (τ) values. It
should be noted that Rf (τ) takes into account all signal
components in the considered
frequency band. They are related to the rotation frequency, the
gear meshing frequency
and its superharmonics as well as components from magneto-electric
phenomena
(magnetostriction). In the case of the cardboard bender (Fig. 5 c),
a non-stationarity in the
amplitude terms is mainly visible. This is due to the cyclical
activation of this subassembly
(natural frequency around 75 Hz, see Fig. 2). The vibration
acceleration signal recorded
on the punch (Fig. 5 b) is non-stationary in the frequency terms,
which is indicated by the
large variability of Rf (τ) with only slight changes in
arms(τ).
The analysis of changes in the averaged values of f, i rms as a
function of machine
capacity is important at the stage of testing the prototype of the
former. One can identify
subassemblies for which increased capacity changes the nature and
intensity of vibrations.
This type of information can be premises for optimizing the machine
design.
a) p.4
b) p.5
c) p.9
Figure 6. The change in the position of the result of the
parameterisation of vibration
accelerations on the Rf – arms plane resulting from the
increase
in machine capacity from 15 to 30 cycles per minute (cpm)
a) gearmotor base b) punch guide c) cardboard bender [4]
0
1
2
3
4
5
f [Hz]
rms [m/s2]
15 cpm
39 cpm
rms [m/s2]
15 cpm
f [Hz]
rms [m/s2]
15 cpm
Vibrations in Physical Systems 2020, 31, 2020112 (9 of 10)
Based on the images in Figure 6, it can be stated that an increase
of the capacity of the
former:
- has no significant effect on the vibration generated by the punch
gearmotor (Fig. 6a),
- causes an increase in the vibration intensity of the punch guide
and slightly changes the
spectral composition (shift of f towards higher frequencies, Fig.
6b)
- causes an increase in the vibration intensity of the cardboard
bender and a decrease in
f due to the greater energy share of the subassembly’s own
frequency. This is due to
the shortening of the time interval between successive excitations
of the subassembly to
vibrate.
5. Conclusions
The graphic representation of the Rf(τ), arms(τ) results on the Rf
– arms plane in the
short-time terms gives information about the nature of signal
instability in both amplitude
and frequency terms as well as the intensity of vibrations of
individual subassemblies.
The averaged values f, and rms obtained for a new machine should be
treated as
reference values in the vibration monitoring process. Changing the
location of the point
specified by f, and rms on the Rf – arms plane during operation
will be a signal of a change
in the technical condition of the monitored subassembly. The way
the results are displayed
enables simultaneous monitoring of many subassemblies of the
machine. The proposed
method can be used to monitor complex machines with many
subassemblies and drive
units with different kinds of working movement.
This method can be used at the stage of machine prototypes testing
as an effective tool
for detection of subassemblies that increase vibroactivity along
with the increase of
machine capacity.
Acknowledgment
Work carried out as part of the NCBR project: POIR.01.01.01-00-0224
/ 17; Modular
production system of SRP (shelf ready packaging) type – PROTIM Sp.
z o. o. Pozna.
References
1. ISO 10816-1:1995, Mechanical vibration – Evaluation of machine
vibration by
measurements on non-rotating parts – Part 1: General
guidelines.
2. ISO 20816-1:2016, Mechanical vibration – Measurement and
evaluation of machine
vibration – Part 1: General guidelines.
3. P. Czop, W. Staszewski, A. Jaboski, Parametric early warning
diagnostic method
for rotating machinery diagnostics, Diagnostyka, 17(4) (2016) 49 –
58.
4. R. Barczewski, B. Jakubek, Project: Modular SRP packaging
production system, :task
1: Vibration and noise studies, Technical Report No r475_2019,
Poznan University
of Technology, 2019.
5. S. Qian, D. Chen, Joint time frequency analysis: methods and
applications, Prentice-
Hall, Upper Shaddle River, 1996.
6. C. Cempel, Diagnostically oriented measure of vibroacoustical
processes, Journal of
Sound and Vibration, 73(4) (1980) 574 – 561.
Vibrations in Physical Systems 2020, 31, 2020112 (10 of 10)
7. P. Deuszkiewicz, D. Górnicka, Rice Frequency as a Measure of
Damage to the
Combustion Engine Valve, Przegld Mechaniczny, 4 (2009) (in
Polish).
8. F. Inoue, E. Outa, K. Tajima, T. Machiyama, An Experimental
Study on Control Valve
Cavitation (1st Report Diagnostics of Cavitation Vibration at High
Pressure
Reduction and at Low Valve Opening), Transactions of the Japan
Society of
Mechanical Engineers Series B, 53(485) (1987) 127 – 137.
9. C. Cempel, Principles of vibroacoustical diagnostics of
machines, WNT, Warszawa
1982 (in Polish)
10. R. Barczewski, Short time Rice frequency analysis (STRFA) a
method of the time
variant vibration signal analysis, Vibrations in Physical Systems,
22 (2006) 79 – 82.
11. R. Barczewski, Diagnostic oriented methods of short time
processing of vibroacoustic
signals, Publishing House of Poznan University of Technology, Pozna
2013
(in Polish).
12. J. Bendat, A. Piersol, Random data: Analysis and Measurement
Procedures, John
Wiley & Sons, New York 1986,
13. M. Barczewski, R. Barczewski, T. Sterzynski, Dynamic pressure
analysis as a tool
for determination of sharkskin instability by extrusion of molten
polymers, Journal of
Polymer Engineering, 32 (2012),
doi:10.1515/polyeng-2011-0157.
14. T. Plazenet, T. Boileau, C. Caironi, B. Nahid-Mobarakeh, Signal
processing tools for
non-stationary signals detection, 2018 IEEE International
Conference on Industrial
Technology (ICIT), Lyon, 2018, 1849 – 1853.